Data-driven revolution of enzyme catalysis from the perspective of reactions, pathways, and enzymes

Enzyme catalysis has emerged as a powerful tool for accelerating the discovery of new synthesis pathways for a diverse array of compounds, impacting drug discovery, agriculture, and the food industry. In recent years, with the continuous accumulation of experimental data, data-driven methodologies have allowed researchers to explore a multitude of biotransformation possibilities with increased accuracy, efficiency, and diversity. This review primarily focuses on data-driven approaches in modeling enzyme catalysis across three hierarchical levels: the reaction level, pathway level, and enzyme level. We highlight essential methods, from the prediction of single-step reactions to the expansion and evaluation of complete pathways and the optimization and design of enzymes with specific catalytic functions. Finally, we summarize the unmet challenges and present future perspectives for the development of modeling enzyme catalysis. We anticipate that this review will deepen the understanding of enzyme catalysis while stimulating new applications and novel endeavors.